Augmenting Collaboration

This project is involved with improving the quality of collaborative learning by means of promoting self and group regulation. This is to be achieved by providing the collaborating participants with real-time feedback on certain aspects of the conversation they are having such as turn-taking patterns and levels of participation, as well as the manner in which each group member is participating. We aim to study the effects of such a table on collaborative learning by conducting experiments and user-studies and following both a qualitative and quantitative approach in analysis of the results.

First generation Reflect tables

Reflect

Reflect is a luminous board embedded in a table. It monitors conversations via microphones and shows a visualisation of the current state of the conversation using arrays of color LEDs. Reflect is intended to act as public shared memory that describes recent and past moves in the conversation. Its display is visible by all participants. Reflect functions like a mirror. It displays some aspects of the conversation to the users but is not intended to give them precise suggestions on what to do next. The feedback provided is peripheral, not requiring an important cognitive load in order not to reduce the ability of the participants to engage in the conversation. In the same line, the system does not require participants to wear any special equipment or to behave in any way that would seem unnatural to them. Microphones are embedded in the board, not attached to the participants via headgears or tie-clips. Reflect is an example of what we call a ``mid-tech'' approach based on simple, cheap and robust pieces of technology. We have deliberately chosen not to equip the table with a large and expensive hi-resolution tactile display. Our initial assumption is that highly detailed feedback is likely to be counter-productive for enhancing and enriching conversations.

First User Study Complete

Having constructed a working prototype of Reflect, we started a user study that would allow us to observe the effect the table can have on collaborating groups. The user-study was designed to test two main hypotheses, in addition to collecting data for further exploratory analysis. The first hypothesis is that students regulate their levels of participation better when these levels are displayed in a shared space. The second hypothesis is that students become more aware of how much they speak when levels of participation are displayed. The study we conducted included 72 participants from first, second and third year undergraduate classes at the University of Lausanne and the EPFL. The students were asked to solve a task together and some crucial information needed to solve the task was given to individual members and were not shared. The result was a collaborative situation in which every student's participation was required to correctly solve the problem at hand. The study partially confirmed both hypotheses. The first hypothesis was confirmed under certain conditions, and the second was confirmed in general. Details on the experiment and the results can be found in a paper published in the European Conference on Technology Enhanced Learning, "Reflect: An Interactive Table for Regulating Face-to-Face Collaborative Learning." Following the formal verification of the hypotheses, a more exploratory approach to the analysis has started, in which we are using the audio and video recordings of the experiments as well as generated logs in order to get deeper insight into, among other things, how and why group regulation may occur.

Second Generation Reflect Table

Ongoing Work

In parallel to the analysis of the experimental results, we are now working on two other tracks for the Reflect table. The first is by using the 16-hour corpus of recorded audio to build a machine learning system capable of detecting voice patterns that can help us understand how the participants collaborate. This multi-step process will start by labeling the corpus with interesting episodes. The other track concerns running a deeper study on the usability and effect of the Reflect table in a more ecologically valid context, i.e. outside the limitation of controlled lab experiments. For that we have developed, and are currently assembling several new prototypes that are similar to the current working one but that address some of its issues and carry a much lower price tag. These will be made available in different locations for long-term use. The tables will gather data on their usage which will allow us to get a better idea on how the table is used "in the wild" and on the effect of using the table for prolonged periods of time.

Relevant Publications